人脸识别因其在安全验证系统、信用卡验证、医学、档案管理、视频会议、人机交互、系统公安(罪犯识别等)等方面的巨大应用前景而越来越成为当前模式识别和人工智能领域的一个研究热点。
人脸识别的关键就是人脸图像预处理,预处理效果的好坏直接关系着人脸识别的结果。本文研究了一个基于PCA人脸识别的人脸图像预处理方法,采用PCA方法就是利用K-L变换和SVD得到正交基,然后根据主特征提取的原理,选取较大特征值对应的基向量。目前算法仅仅针对单人正面的图像,有很大的局限性。
本文所采用的方法,除了对算法的优化外,更加注重图像预处理的效果。
关键字:人脸识别;人脸预处理;光照补偿;K-L变换;面部特征定位
Research and Implementation Of Image Pre-processing Algorithm Based on Face Recognition
Abstract
Face recognition is important in surveillance and security, telecommunications, digital libraries , video meeting, and human-computer intelligent interactions. It has been a research focus of pattern recognition and artificial intelligence.
The important of face recognition is face image preprocessing. The pretreatment directly effect the results of the face recognition. In this paper, we study and implement a face image preprocessing method based on PCA. PCA use K-L transform and SVD to get orthogonal basis, and then according to the principles of the main feature extraction to select the larger eigenvalues of the corresponding vector-based. The current algorithm is only for the single positive face image and it have limitation.
The method in this paper uses algorithm optimization and the effects of image preprocessing to achieve the performance of the algorithm.
Keyword:Face recognition;Face pretreatment;Light compensating;K-L transform;Characteristics of position